论文部分内容阅读
为了更真实地模拟现实知识网络的成长过程,构造了一类基于局域连接机制下的知识网络生成模型.利用统计物理学中的平均场方法,给出了知识网络度分布的理论预测,并运用Matlab仿真进行了验证:当局域世界规模很小时,网络度的分布函数近似服从无标度分布,当局域世界规模不是很小时,网络度的分布会从纯粹的无标度状态变化成尾状物服从指数分布的近似无标度状态,且无标度指数随着可调参数增加而增加,随着新增边数的增加而减少.最后,比较了在局域连接机制和全局连接机制下生成的知识网络的一些知识指标.从长期来看,与全局机制相比,局域机制会导致网络平均知识水平增长缓慢,而且网络同质化现象严重.
In order to simulate the growth process of real knowledge network more faithfully, a class of knowledge network generation model based on the local area connection mechanism is constructed. By using the average field method in statistical physics, the theoretical prediction of knowledge network degree distribution is given. It is verified by Matlab simulation that when the size of the local area is small, the distribution function of network degree follows a scale-free distribution. When the size of the local area world is not very small, the distribution of network degree changes from a pure scale-free state to a caudate The object obeys the approximately scale-free state of exponential distribution, and the scale-free index increases with the increase of the adjustable parameters and decreases with the increase of the number of new edges.Finally, we compare the mechanism of local connection and global connection In the long run, compared with the global mechanism, the local mechanism will cause the average knowledge level of the network to grow slowly and the network homogeneity will be serious.